Ensuring access to clean water for generations to come will involve developing novel ap- proaches to determining the safety and composition of potable water that are practical and afford- able. Arsenic, mercury, and cadmium are three of the top priorities among hazardous substances commonly found at Superfund sites, as they are linked to health problems in people exposed to them in drinking water, yet the current real-time monitoring methods for these and other contam- inants are either extremely costly or nonexistent, making it dif?cult to monitor water quality with high spatial or temporal resolution. QBiSci is developing a biosensor that uses synthetic micro- bial sensor strains that ?uoresce in response to speci?c toxins to continuously monitor water for contamination. The platform will substantially improve upon currently available technologies for toxin detection, making monitoring more affordable, continuous, and ?eld-deployable. Speci?c Aim 1: To fully characterize three synthetic E. coli strains that speci?cally detect ar- senic, mercury, and cadmium in a continuous water stream. For a real-time sensor to be maxi- mally effective, it must be able to report accurate toxin concentrations in real-time. Focusing on three of the highest priority contaminants as a proof of feasibility, comprehensive data will be acquired to train a machine learning algorithm to be able classify real-world samples in real-time. Speci?c Aim 2: To develop and train a classi?cation algorithm to recognize the type and amount of each contaminant present in a continuous water stream. The ability to analyze and interpret data in real-time from a constantly ?uctuating water source will require an extensive classi?cation train- ing effort. QBiSci's existing machine learning framework will be trained and tested using many contamination induction scenarios, ranging from sudden pulses to subtly varying concentrations. Speci?c Aim 3: To develop a micro?uidic cartridge system that reduces device complexity and enables sensor deployment with minimal intervention. QBiSci will develop a swappable car- tridge system using devices that are pre-loaded with biologically-stable strains and can simply be ?plugged in? to the sensor platform to achieve repeatable results in a user-friendly manner. The development of a method for thermoplastic device fabrication will enable the more precise connections required for a cartridge clamping system that will require little operational expertise. A successful outcome of this proposal will lead to a biosensor capable of real-time quanti?ca- tion of arsenic, mercury, and cadmium in a continuous water input. A future Phase II proposal would focus on real-world performance evaluations of our sensors via deployment in areas of con- cern and comparison of our results to standard techniques as well as an expansion of the platform to detect other contaminants quantitatively and continuously. 1